The polypeptide fold of the globular domain of histone H5 in solution. A study using nuclear magnetic resonance, distance geometry and restrained molecular dynamics.

Abstract

The polypeptide fold of the 79-residue globular domain of chicken histone H5 (GH5) in solution has been determined by the combined use of distance geometry and restrained molecular dynamics calculations. The structure determination is based on 307 approximate interproton distance restraints derived from n.m.r. measurements. The structure is composed of a core made up of residues 3-18, 23-34, 37-60 and 71-79, and two loops comprising residues 19-22 and 61-70. The structure of the core is well defined with an average backbone atomic r.m.s. difference of 2.3 0.3 A between the final eight converged restrained dynamics structures and the mean structure obtained by averaging their coordinates best fitted to the core residues. The two loops are also well defined locally but their orientation with respect to the core could not be determined as no long range (li-jl > 5) protonproton contacts could be observed between the loop and core residues in the two-dimensional nuclear Overhauser enhancement spectra. The structure of the core is dominated by three helices and has a similar fold to the Cterminal DNA binding domain of the cAMP receptor protein.

Determination of three‐dimensional structures of proteins from interproton distance data by hybrid distance geometry‐dynamical simulated annealing calculations

“…Although the atomic RMS distribution for GH5 is increased by -50% after full DISGEO calculations, it is doubled by dynamical simulated annealing (table 2). A similar increase in atomic RMS distribution is also obtained after subjecting the (DG) structures to restrained molecular dynamics refinement [21]. In the case of crambin, however, the overall structure is welldefined both locally and globally by the data.…”

“…The atomic RMS distributions and shifts of the structures before and after dynamical simulated annealing are summarised in table 2, and the backbone atomic RMS distribution of the individual SA structures about their mean as a function of residue number if shown in fig.2. GH5 contains an approximately equal mixture of welland poorly defined regions [21]. The atomic RMS distributions of the GH5 substructures is approximately the same as those in the case of crambin (table 2), despite the fact that the latter is only half the size of the former and much better defined by the data [8,21].…”

“…The same sets of NOE distances were employed as in our previous calculations [8,21]. In the case of crambin the NOE data set consisted of 240 interproton distances derived from the crystal structure [30]; for GHS it comprised the 307 interproton distances derived from NOE measurements [21]. The distances were classified into three distance ranges, 1.8-2.7, 1.8-3.3 and 1.8-5.0 A, corresponding to strong, medium and weak NOES [3 11.…”

“…Distances to methyl, methylene and aromatic ring protons that were not assigned stereospecifically were calculated with respect to the average position of these protons and the upper limits of the corresponding restraints were corrected appropriately as described in [32]. In the a Notation of structures: (Sub), converged substructures obtained from phase 2 of DISGEO, after best fitting of residues and subsequent unrestrained minimisation (see text); (SA), converged structures produced by dynamical simulated annealing starting from DISGEO substructures; (DC), final converged structures obtained with DISGEO alone (from --- [33], for crambin; from [21] for GH5); SA, DC, Sub, mean structures obtained by averaging over the coordinates of individual SA, DC, Sub structures, respectively; (m)r, structure obtained by restrained minimisation of the mean s structure. There are 9 (Sub) and (SA) structures for crambin, and 10 for GHS; and 7 (DC) structures for crambin and 6 for GH5 b Interproton distance deviations calculated with respect to upper and lower limits of restraints.…”

“…GH5 contains an approximately equal mixture of welland poorly defined regions [21]. The atomic RMS distributions of the GH5 substructures is approximately the same as those in the case of crambin (table 2), despite the fact that the latter is only half the size of the former and much better defined by the data [8,21]. Thus, as noted in [24], the relatively inefficient sampling of the available conformafional space by metric matrix distance geometry calculations is already present in the substructuregenerating phase.…”